Page 43 - Read Online

Page 307 Brenac et al. Art Int Surg 2024;4:296-315 https://dx.doi.org/10.20517/ais.2024.49

performed significantly better
Hoogendam Predicting clinically relevant 2022 Neurosurgery (1) Develop a prediction model that estimates the (1) A gradient boosting machine model with 5 predictors
[48]
et al. patient-reported symptom probability of clinically relevant symptom was identified as the best balance between
improvement after carpal tunnel improvement 6 months after CTR discriminative ability and simplicity, achieving an AUC of
release: a machine learning (2) Evaluate the model’s discriminative ability and 0.723 in the holdout data set
approach calibration using various ML techniques and apply (2) The model demonstrated good calibration, with a
it to support shared decision making for patients sensitivity of 0.77, specificity of 0.55, positive predictive
considering CTR value of 0.50, and negative predictive value of 0.81
(3) The prediction model, which uses 5 patient-reported
predictors (18 questions), has reasonable discriminative
ability and good calibration, and is available online to
assist in shared decision making for patients considering
CTR
[49]
Loos et al. Machine learning can be used to 2022 Clinical (1) To develop and validate prediction models for (1) The random forest model for pain prediction showed
predict function but not pain after Orthopaedics and clinically important improvement in pain and hand poor performance with an AUC of 0.59 and poor
surgery for thumb carpometacarpal Related Research function 12 months after surgery for thumb calibration
osteoarthritis carpometacarpal osteoarthritis (2) The gradient boosting machine model for hand
(2) Assess the performance of various predictive function improvement had a good AUC of 0.74 and good
models using logistic regression, random forests, calibration, using only the baseline hand function score
and gradient boosting machines to support as a predictor
preoperative decision making (3) A web application is available for the hand function
model, which could aid in clinical decision making,
though the pain prediction model is not yet suitable for
clinical use
[50]
Wound healing and Kim et al. Predicting the severity of 2023 Scientific Reports (1) Develop and evaluate an AI model using (1) The AI model reached a high level of accuracy (ROC-
burn surgery postoperative scars using artificial images and clinical data to predict the severity of AUC 0.931 for images alone, 0.938 combined with
intelligence based on images and postoperative scars clinical data)
clinical data (2) Compare the performance of this AI model to (2) The model also performed at a comparable level to
that of dermatologists that of 16 dermatologists
[51]
Squiers et al. Machine learning analysis of 2022 Journal of Vascular (1) Develop a ML algorithm using multispectral (1) The ML algorithm had high sensitivity (91%) and
multispectral imaging and clinical Surgery imaging data and clinical risk factors to predict specificity (86%) for prediction of non-healing
risk factors to predict amputation amputation wound healing and reduce the need for amputation sites
wound healing reoperation (2) ML algorithms could reduce reoperation rates,
improve healing outcomes, and potentially decrease
costs and patient length of stay
[52]
Robb Potential for machine learning in 2022 Journal of Burn (1) Explore the potential implementation of various (1) The use of ML in burns holds the potential to improve
burn care Care & Research ML methods (such as linear and logistic prevention, burns assessment, mortality predictions, and
regression, deep learning, and neural networks) in critical care monitoring
burn care within the NHS in the UK (2) Successful implementation requires investment in
(2) Focus on optimizing care through ML data capture and training
applications in burn assessment (3) ML technology has the potential to improve
diagnostic accuracy, objective decision making, and
resource allocation
[53]
Xue et al. Artificial intelligence - assisted 2022 ACS Applied (1) Explore potential therapeutic agent TSA for (1) TSA via microneedle patch reduces inflammation,
bioinformatics, microneedle, and Materials & diabetic wound healing with AI-assisted promotes tissue regeneration, and inhibits HDAC4 in
diabetic wound healing: a “new Interfaces bioinformatics diabetic wound healing
deal” of an old drug (2) Investigate the effectiveness of TSA in (2) This approach offers a minimally invasive and safe